JLM, BJMdW, LMP, CMR, PMR, and SG conceived and designed the experiments. JLM, BJMdW, and CMR performed the experiments. JLM, BJMdW, CMR, PMR, and SG analyzed the data. LMP, CMR, RAD contributed reagents/materials/analysis tools. JLM, BJMdW, LMP, CMR, and SG wrote the paper.
¤ Current address: Dublin-Oxford Glycobiology Laboratory, National Institute for Bioprocessing Research and Training, University College Dublin, Dublin 4, Ireland
The authors have declared that no competing interests exist.
Macrophages (MØ) and mononuclear phagocytes are major targets of infection by dengue virus (DV), a mosquito-borne flavivirus that can cause haemorrhagic fever in humans. To our knowledge, we show for the first time that the MØ mannose receptor (MR) binds to all four serotypes of DV and specifically to the envelope glycoprotein. Glycan analysis, ELISA, and blot overlay assays demonstrate that MR binds via its carbohydrate recognition domains to mosquito and human cell–produced DV antigen. This binding is abrogated by deglycosylation of the DV envelope glycoprotein. Surface expression of recombinant MR on NIH3T3 cells confers DV binding. Furthermore, DV infection of primary human MØ can be blocked by anti-MR antibodies. MR is a prototypic marker of alternatively activated MØ, and pre-treatment of human monocytes or MØ with type 2 cytokines (IL-4 or IL-13) enhances their susceptibility to productive DV infection. Our findings indicate a new functional role for the MR in DV infection.
Dengue disease and its severe manifestations are a growing public health concern, with a third to half the world's population living in dengue-endemic areas. In recent years there have been significant advances in understanding dengue virus (DV) interactions with target cells such as macrophages, dendritic cells, hepatocytes, and endothelial cells. Interaction with and infection of these cells leads to the production of new virions as well as immune mediators, which can shape the course of the subsequent immune response. The vascular leakage associated with dengue haemorrhagic fever is believed to be immune mediated. Our work on the interaction of DV with human macrophages has led to two major findings; first, we have identified that the macrophage mannose receptor is important for mediating the infection of human macrophages by DV, and second, that the type 2 cytokines IL-4 and IL-13 enhance macrophage susceptibility to DV infection. DV–receptor interactions are of critical importance for understanding not only the mechanisms of entry, but also the biology of infection and the pathogenesis. Understanding the immunopathogenesis of dengue disease is crucial to the development of both a safe dengue vaccine and therapeutic inhibitors of early DV replication.
Dengue is the most prevalent mosquito-borne viral disease worldwide and in the past 40 years has undergone a global resurgence such that almost half the world's population are currently living at risk in dengue-endemic areas [
MR is a multi-domain protein that is composed of a cysteine-rich (CR) domain which has lectin activity and binds to sulphated sugars, a fibronectin type-II (FNII) domain that mediates binding to collagen [
While the immune response to viruses is classically described as Th1 mediated, the literature in the case of DV suggests that this may not be absolute. IgE (characteristic of a Th2 environment) has recently been shown to be elevated in the acute stages of DV infection [
MØ are profoundly influenced by the cytokine profile in their immediate environment. Functionally diverse subsets of alternatively or classically activated mononuclear phagocytes can develop in an immune response. Exposure of MØ to IL-4 or IL-13 elicits an ‘alternate type of activation’, as opposed to the classical activation induced by IFNγ [
In this study we show that MR binds to DV grown in mosquito cells and to recombinant mammalian cell–produced DV envelope glycoprotein. A recombinant MR fusion protein (CRD4–7-Fc) was shown to recognize DV envelope (E) protein in ELISA and blot overlays, and binding was inhibited by mannose, fucose and EDTA. The presence of MR on transfected cells is sufficient to confer DV binding. DV infection of MØ was blocked by antibodies against the human MR suggesting that it is a novel functional receptor contributing to DV infection. We also show that pre-treatment of primary human monocytes with Th2 cytokines (IL-4/IL-13), which upregulate MR expression, increases their susceptibility to DV infection in vitro. Better understanding of receptor/s and entry pathways mediating infection in humans could be crucial to the design and safety of a dengue vaccine.
The ability of MR to bind DV antigen produced in mosquito (C6/36) and human (293T) cells was examined. ELISA wells were coated with semi-purified C6/36-grown DV2 or recombinant soluble E (sE) protein produced in the endothelial kidney cell line 293T (see below for characterisation of this reagent) and probed with the entire extracellular region of the murine MR expressed with an HA tag or recombinant truncated forms of the murine MR with human Fc tags. MR-HA bound to purified mosquito cell–derived DV2 (
(A) Binding of MR extracellular domain (MR-HA) and chimeric MR fusion protein constructs to mosquito cell–derived (C6/36) NGC strain DV2 and (B) soluble E protein (sE) produced in a human cell line, detected by ELISA. Closed bars, DV antigen; open bars, uncoated wells. The MR-HA protein was detected with an anti-murine MR antibody (MR5D3). Binding of the Fc fusion proteins to the DV/sE was detected with an anti-Fc antibody. (C) Inhibition of CRD4–7-Fc binding to mosquito cell–derived semi-purified DV2 and (D) sE, detected by ELISA as above. (E) Binding of CRD4–7-Fc to all four serotypes of DV, detected with an anti-Fc antibody. Strains examined were mosquito cell (C6/36)–derived virus of DV1 (Hawaii), DV2 (NGC), DV3 (H-87) and DV4 (H-241), and Vero cell–grown DV2 (16681). Differential binding to the serotypes may reflect coating levels, as indicated by (F) binding of rabbit anti-DV1–4 antibody. Normal rabbit immunoglobulin (RIg) was included as a control and both were detected with an anti-rabbit antibody. The low levels of CRD4–7-Fc binding to C6/36-grown DV2 in this panel reflect a shorter development time than in the other panels. Data are expressed as mean and SD of triplicate wells. Representative data from two to five independent experiments are shown. Recombinant fusion proteins contain the cysteine-rich (CR) domain, fibronectin type-II (FNII) domain, and various carbohydrate-recognition domains (CRD) of the MR.
We extended the study by investigating the binding of CRD4–7-Fc to mosquito cell–derived virus of the other 3 DV serotypes (DV1, DV3 and DV4) and to mammalian (Vero) cell–grown DV2 by ELISA. CRD4–7-Fc bound to all four serotypes of DV in a dose-dependent manner (
The specificity of MR CRD4–7-Fc binding to DV sE was further examined by blot overlay. CRD4–7 bound exclusively to a single band that migrated at 52 kDa (
sE and PNGaseF-treated sE were resolved by SDS-PAGE using 10% gels, transferred to nitrocellulose membranes, and (A) probed with MR CRD4–7-Fc. (B) Blots were subsequently stripped and reprobed with the anti-E protein antibody, 3H5.
Given the interaction of DV with MR described above, it was important to characterise the glycans on the human cell–produced sE, especially since we are unaware of any similar analysis in the literature. This reagent is valuable, as its glycan modifications may more closely resemble the patterns found on viral particles produced during infection in the human host compared with baculovirus and
(A) SDS-PAGE of sE protein preparation resolved on 10% gel and stained with Coomassie Brilliant Blue. Lane 1, sE preparation as eluted from NiNTA-agarose. Lane 2, PNGaseF-treated sE. Lane 3, PNGaseF alone.
(B) Western blot of sE before (Lane 1) and after treatment with PNGaseF (Lane 2) or EndoH (Lane 3), resolved under reducing conditions by 10% SDS-PAGE, and probed with the anti-DV E monoclonal antibody, 3H5.
(C) NP HPLC chromatograms of the complete pool of 2AB labelled glycans of sE glycoprotein, together with sequential digestions. a) Complete glycan pool, undigested, with structural representation of A2G2S2. b) Glycans digested with
Key: A2, biantennary; G, galactose; F, fucose; S, sialic acid. Filled square,
To further evaluate MR as a potential DV receptor, we examined binding of DV to human MR-transfected 3T3 cells (3T3.hMR). As DC-SIGN has previously been shown to be an important attachment receptor for DV, DV binding to 3T3.hMR cells was compared with binding to 3T3 cells transfected with DC-SIGN. Initially we confirmed expression of the respective receptors on the 3T3 cell surface (
The expression levels of (A) MR and (B) DC-SIGN on cells transfected with MR (3T3.hMR), DC-SIGN (3T3.DC-SIGN), or vector only (3T3) were assessed by flow cytometery. Receptor expression was detected with 15–2 (anti-hMR) or 120507 (anti-DC-SIGN) antibodies (pale and dark blue lines). Matched isotype control (purple and green lines) and secondary antibody only (brown and red lines) staining is included. Binding of NGC DV2 to cells transfected with (C) vector only (3T3), (D) MR (3T3.hMR), and (E) DC-SIGN (3T3.DC-SIGN) for 90 min on ice was detected by flow cytometry. The histograms show the binding of anti-DV2 antibody (3H5; brown and green lines) and isotype matched control antibody (blue and red lines). The relative fluorescence intensity was measured by FACSCalibur analysis and the data are normalised and presented as percent of maximum. Representative data from one of two independent experiments are shown.
A human primary cell culture assay system was established in which we examined the functional role of MR and the effects of cytokines on the susceptibility of mononuclear phagocytes to DV infection. Monocytes were purified from human PBMC fractions and cultured for 2 or 7 d to prepare monocyte-derived MØ (MDMØ) or differentiated into monocyte-derived dendritic cells (MDDC) prior to infection. The percentage of cells infected was quantified by microscopy by staining nuclei with DAPI and viral antigen with the anti-DV2 E protein monoclonal antibody, 3H5. MDDC were more susceptible to infection by DV (percent infected DC: 12.3% +/− 7.1) compared with either 2 or 7 d differentiated MDMØ in the absence of added cytokine (percent infected 2 d MDMØ: 1.8% +/− 0.7; 7 day MDMØ: 1.1% +/− 0.3) from the same donor (3 donors), using a multiplicity of infection of 0.4. The presence of DV non-structural protein in infected 2 d MDMØ using anti-NS1 monoclonal antibodies (obtained from Eva Harris) suggested that active viral replication and de novo viral protein production, rather than mere uptake of viral antigen, was occurring (data not shown). Plaque assays on supernatants from infected primary 2 or 7 d MDMØ cell cultures confirmed the occurrence of productive infection, with viral titres increasing over time and reaching 104 pfu/ml in cell supernatants 2 d after infection of 6 × 106 cells.
Considering that MDDC, which were grown in an IL-4/GM-CSF cytokine cocktail, were infected to a higher degree than the MDMØ, various cytokines were tested for their ability to alter the susceptibility of MDMØ to DV infection by pre-incubation with monocytes for 48 h prior to DV infection (see
Monocytes isolated from human peripheral blood were either treated with IL-4 or IL-13 or were left untreated for 2 d (day 2 MDMØ) prior to infection with dengue virus. Two days following infection, permeabilized cells were stained with antibody to dengue E protein (green) and nuclei stained with DAPI (blue).
(A) Low power image. (B) Single DV-infected cell confocal image showing DV protein distributed throughout the cytoplasm. Image represents a single
Neither the age of the cells nor the length of treatment altered the enhancing effects of IL-4. Enhanced susceptibility to DV infection was seen when monocytes were allowed to differentiate into MØ over 7 d, and then treated with IL-4 for 48 h prior to DV infection (day 9 MDMØ,
The interaction of DV with primary cells appears to be multifactorial, as we found variability between donors and receptor expression level. Surface expression of both MR and DC-SIGN is upregulated on MDMØ by IL-4 treatment (
The effects of 2 d IL-4 treatment (25ng/ml) of human monocytes on surface expression levels of (A) MR and (B) DC-SIGN were assessed by flow cytometery. Receptor expression was detected with 15–2 (anti-hMR) or 120507 (anti-DC-SIGN) antibodies (black lines) on untreated cells (open histogram) and IL-4-treated cells (filled histogram). Matched isotype control (grey lines) staining is included. The relative fluorescence intensity was measured by FACSCalibur analysis and the data are normalised and presented as percent of maximum. Representative data from one of eight donors are shown.
The functional role of MR in DV infection of primary human MØ was investigated using a polyclonal anti-MR antibody to block infection. This was examined in the IL-4-treated MDMØ since these cells showed the highest rate of DV infection. Anti-MR antibody significantly blocked DV infection of IL-4-treated MDMØ (p = 0.008) in all donors tested (6 donors;
(A) IL-4-treated human monocytes were incubated in triplicate wells with medium alone (no block), 2mg/ml mannan, titrations of goat anti-human MR antiserum, or with normal goat serum (NGS). Treated cells were infected with mosquito cell–grown 16681 DV2 at a multiplicity of infection of 0.5 in the presence of these inhibitors, incubated for 48 h and fixed.
(B) Monocytes were treated in triplicate wells, as above, including blocking with monoclonal antibodies specific for DC-SIGN (120507), DC-SIGNR (120604), and both DC-SIGN and DC-SIGNR (DC28 and 120612) or isotype controls (all at 5ug/ml) prior to infection with mosquito cell–grown 16681 DV2 at a multiplicity of infection of 0.04 in the presence of these inhibitors. Following fixation, cells were immunolabelled with anti-DV E protein monoclonal antibody 3H5 and counted using a fluorescent microscope, and percent of cells infected calculated. Data are expressed as mean and SD of triplicate wells. Representative data from one of multiple donors are shown.
(C) The titre of infectious virus in the cell supernatant at 48 h post infection was determined by plaque assay. Due to variation between donors, the titre at 48 h in the absence of block was normalised to 100. Each data point is the average of three infected wells, each plaqued in triplicate, and the results from six donors are shown in this graph (mean represented by a bar). The p value was calculated by unpaired, two-tailed
Mannan, which blocks MR, DC-SIGN and other receptors with specificity for mannose, also blocked DV infection of MDMØ (
We have shown for the first time that MR is a functional receptor for DV infection of human MØ. Binding of the MR to DV surface glycoproteins was mediated via the lectin activity of the CRD binding to glycans on the DV E protein. Gain of function binding data showed that surface expression of human MR on 3T3 cells was sufficient to confer DV binding. Antibodies specific for the MR significantly blocked both infection of MDMØ and the production of infectious virus in these cells. FACS analysis showed surface MR expression increased over 4-fold following IL-4 treatment of monocytes, corresponding with a similar fold increase in percent infected cells. Thus, the MR provides a potential link explaining the increase in MØ permissiveness to DV when stimulated by IL-4 or IL-13. We hypothesise that the MR may play a role in at least one of the stages of DV infection of human MØ. The first stage of virus entry into a cell is attachment, and the mechanism by which MR enhances the efficiency of DV entry could be by increasing virus attachment, as suggested for DC-SIGN and DV. As the MR can be internalized by macropinocytosis, pinocytosis, receptor-mediated endocytosis and phagocytosis, its role could also be in increasing the rate of DV internalisation, the second stage of virus entry. Analogous to the proposed mechanism of Fc-receptor enhancement of DV-antibody complex attachment/uptake in antibody-dependent enhancement, the presence of receptors such as MR or DC-SIGN, which enhance virus attachment/entry, may play a significant role in vivo.
Anti-DC-SIGN monoclonal antibodies were also able to block DV infection of MDMØ, as has been seen previously in MDDC. The anti-DC-SIGN antibodies that blocked DV infection of IL-4 treated MDMØ to the greatest degree (DC28 and 120612) are known to cross react with DC-SIGNR. We hypothesise that the different specificities of the monoclonal antibodies may explain why some but not all block DV infection of MØ. These findings corroborate data by Tassaneetrithep et al. [
MR is ideally poised to act as a DV entry receptor given its constitutive recycling to the cell surface and ability to promote ligand internalisation via both endocytic and phagocytic pathways. While DC-SIGN mainly localises to the plasma membrane [
MR and DC-SIGN both contain lectin domains, but differ distinctly in terms of ligand specificity, with MR binding terminal mannose, fucose and N-acetyl glucosamine, and DC-SIGN binding mannose within high-mannose oligosaccharides and fucosylated glycans [
We expanded our study of the interaction of MR with DV by demonstrating binding of CRD4-7-Fc to all four serotypes of DV. Differences in glycosylation between DV serotypes, and more broadly between different flaviviruses, may be relevant for interaction with lectin receptors such as MR and DC-SIGN. We showed that MR can bind in ELISA to Japanese encephalitis virus and tick-borne encephalitis virus, both of which are reported to have glycosylated envelope proteins.
In this report we have shown that the type 2 cytokines IL-4 and IL-13 enhance the susceptibility of MDMØ to DV infection. The mechanisms resulting in increased infection in response to IL-4 and IL-13 are unknown. Analysis of IL-4-treated human monocytes showed that these cells are characterised by the overexpression and enhanced function of several endocytic receptors, including scavenger and C-type lectin receptors [
A number of important conditions result in polarised activation of MØ phenotype. Our findings make it highly relevant to understand the clinical and epidemiological significance of a Th2 environment on DV infection, pathogenesis, and enhancement and in the development of a desirable vaccine response. It will be of great interest to examine the broader context in which dengue pathogenesis occurs by considering the effects on DV disease of settings that induce Th2 cytokines, including co-infection with parasites, the presence of immune complexes and allergy (e.g. asthma). There are few studies into the implications of co-incidence of parasitic infections or allergic disease and dengue infection. Guzman and colleagues showed significantly enhanced replication of DV in PBMC from asthmatic patients compared with controls [
16681 and New Guinea C (NGC) strains of DV2 (both gifts from E. Gould, Oxford Centre for Ecology and Hydrology, UK) were propagated in the
A stable NIH3T3 cell line expressing the human MR (3T3.hMR) was a gift from Gordon Brown (University of Cape Town, South Africa) and Philip Taylor (University of Cardiff, UK), made using the protocol described previously [
Fc chimeric proteins (derived from the murine MR) and the hemagglutinin-tagged form of MR (MR-HA) were prepared by Richard Stillion as described previously [
For the generation of MDMØ and MDDC, human PBMC were isolated from buffy coats (NHS Blood and Transport) by centrifugation over a Ficoll-PaqueTM PLUS (Amersham) gradient, according to standard protocols. Adherent monocytes, isolated as described previously [
Monocytes were treated with recombinant human IL-4 (25ng/ml; Peprotech) or IL-13 (10ng/ml; Peprotech) for 2 d then infected with mosquito cell-grown 16681 DV2 virus. After 1 h the viral supernatant was replaced with cell culture medium without cytokine and the cells incubated for 2 d then fixed with 4% paraformaldehyde. In some experiments the IL-4-treated MDMØ were incubated with various blocking reagents for 40 min at 37°C before the addition of 16681 DV2 virus (in the presence of blocking agent).
96-well Maxi-Sorp plates were coated overnight with DV antigens (50ug/ml C6/36-grown NGC DV2 and 20ug/ml sE, or as specified in
CRD4–7 binding to sE was investigated by blot overlay by running 0.9ug sE protein and 0.9ug sE protein deglycosylated with peptide: N-glycosidase F (New England Biolabs) on 10% SDS-PAGE, and subsequently transferring proteins to Hybond C-Extra nitrocellulose membranes (Pharmacia). Blots were blocked for 1 h in 0.5% skimmed milk powder in TBS containing 10mM CaCl2 (blocking/washing solution). Blots were probed with 1ug/ml MR CRD4–7-Fc in the absence or presence of either 2mM
An open reading frame consisting of the last 20 aa of C-protein, the entire prM protein, and E protein truncated by 96 amino acids at the C-terminus and containing a hexahistidine tag was amplified by PCR from cDNA prepared from DV2 strain 16681. The expression cassette was cloned in the mammalian expression vector pLEX [
Glycans were released from approximately 25ug of recombinant soluble dengue virus E-glycoprotein and labelled by reductive amination with the fluorophore 2-aminobenzamide [
FACS was performed according to conventional protcols at 4°C in the presence of 2mM NaN3. Non-specific binding sites on cells were blocked with PBS containing 5% heat-inactivated rabbit serum, 5% heat-inactivated goat serum, 0.5% BSA and 5mM EDTA (blocking buffer) before the addition of primary antibodies. Surface expressed MR and DC-SIGN was detected using 10ug/ml 15–2 (Serotec) and 120507 (R&D Systems) monoclonal antibodies, respectively, and was compared with an isotype control (Serotec). Surface bound DV was detected using 10ug/ml 3H5. The primary antibodies were detected using Alexafluor 488-conjugated anti-mouse antibody (Molecular Probes) diluted 1:200 in blocking buffer. Cells were fixed with 1% paraformaldehyde in PBS before analysis. Binding was quantified on a FACSCalibur flow cytometer and data from ∼10,000 cells were routinely acquired for each sample. Data were analysed using FlowJo software (Treestar). Percent of max represents the number of events normalised according to FlowJo algorithms. Fold increase in receptor expression was measured by geometric mean fluorescent intensity of specific receptor antibody staining for IL-4-treated cells divided by untreated cells.
Cells suspensions were prepared by scraping cells to preserve receptor expression at the cell surface. Mosquito cell-grown NGC DV2 (1.5 × 106 pfu/ml) or media was incubated with cells (4 × 106) at a multiplicity of 0.35 infectious virions per cell for 80 min on ice. Unbound virus was washed away with cold media, the cells were fixed with 1% paraformaldehyde in PBS, and surface bound virus detected with anti-DV E-protein antibody by flow cytometry as described above.
FITC-labelled, mannosylated or galactosylated BSA (5ug/ml; Sigma) was incubated for 90 min at 37°C with primary human MDMØ or 3T3 transfectants plated on tissue culture-treated plastic. For blocking studies cells were pre-incubated with mannan (2mg/ml), normal goat serum (NGS) or goat-anti hMR antibody for 20 min at 37°C. After incubation cells were washed with PBS, harvested using PBS containing 5mM EDTA and lidocaine (4mg/ml) and fixed in 2% paraformaldehyde in PBS. Binding was quantified by a FACSCalibur flow cytometer and analysed using FlowJo software.
Fixed, DV-infected cells were permeabilised with 0.5% Triton-X and stained with 10ug/ml 3H5 followed by an Alexafluor 488-labelled secondary anti-mouse IgG antibody (Molecular Probes) and the nuclei stained with DAPI. Stained coverslips were mounted in DakoCytomation fluorescent mounting medium (Dako), and analyzed using either CCD1 (Axioplan) and CCD camera (Spot) or a METATM confocal microscope linked to LSM 510TM software (Carl Zeiss MicroImaging, Inc.). Confocal images were acquired sequentially using the multitrack configuration of the Zeiss METATM to avoid bleed-through between fluorescence channels, and the appropriate controls with and without primary antibody were performed. Additional image processing was performed using Adobe Photoshop 7. The image is presented as single two-dimensional
Statstics were calculated using GraphPad PRISM (version 2.0; GraphPad Software, San Diego, CA) and Microsoft Excel. Two-tailed Student's
(2.1 MB TIF)
(35 KB RTF)
We acknowledge Alexandre Akoulitchev and Ben Thomas at the Central Proteomics Facility of the Dunn School of Pathology for mass spectrometry analysis, Eva Harris and Jennifer Kyle for monoclonal antibodies to dengue virus and helpful advice, Richard Stillion for preparation of Fc chimeric and HA-tagged mannose receptor proteins, and Alan Barratt for recombinant domain III.